Representation of Interrogative Turns in the DB ‘Intonation of Russian Dialog’ Sandro V. Kodzasov, Anastasija A. Bonch-Osmolovskaja, Leonid. M. Zakharov, Irina. M. Kobozeva Department of Theoretical and Applied Linguistics Moscow State Lomonosov University, Moscow, Russia sancod@philol.msu.ru,abonch@philol.msu.ru, leon@philol.msu.ru, kobozeva@philol.msu.ru Abstract Although semantic and pragmatic functions of intonation in dialogical discourse and its correlation with syntax have been extensively studied from different theoretical perspectives for many languages including Russian still there was no linguistic resource that would store representative dialogical material simultaneously marked for prosodic, syntactic, semantic and pragmatic features. In the report we present the first results of the attempt to construct such a data base called “Intonation of Russian dialogue” on the basis of the cognitive-discursive approach to intonation functions with the help of modern tools of phonetic analysis and demonstrate its interrogative component. 1. Introduction This Modern theoretical and applied linguistics is characterized by a rapid increase of interest in the problems of dialogical interaction of different forms and genres. Intonation plays a tremendous role in a dialog. Recent investigations in prosody focus on the informational and discursive aspects of intonation semantics and on the corresponding prosodic means (see e.g. [1], [2], [3]). Not only tonal configurations, reflecting modal and illocutionary properties of utterances but also accent, phonation and timbre structures and their semantic and discursive purport have become an object of study. In Russian intonology such a complex analyses of intonation is only beginning. We consider the adoption of such an a approach for Russian language material important and productive. But until now there has been no intonation data bases for Russian corresponding to modern standards. That is why the construction of an electronic information system based on the latest achievements in this field is an urgent task. Now such a system named “Intonation of Russian Dialog” is under development at the Department of theoretical and applied linguistics at MSU. We apply cognitive-discursive approach in analyzing intonation functions and use modern means for its phonetic analysis. The data are supposed to be taken from the freely accessible audio-recordings of dialogical texts as well as gathered especially for the system. The DB, implemented in Microsoft Access environment, shall allow to extract and to sort the data both by semantic and phonetic parameters, to hear the pronounced utterances on line and to observe their acoustic characteristics. The process of the DB construction is divided into steps according to dialogical micro-scenarios: question – answer, request – reaction etc. At present only the first version of the “interrogative component” of the system is available. So we shall confine our discussion to this component. For constructing a truly representative DB on Russian interrogative dialogical turns a new comprehensive classification of interrogative sentences (IS) was elaborated. It takes into account the fundamental treatises of IS (theoretical grammars, e.g. [4], and monographs, e. g. [5]) and other papers concerning Russian IS and its scope was verified on the list of 500 interrogative turns, selected from TV talk shows. The classification is multi-aspectual in that it reflects illocutionary, informational, syntactic and lexical properties of IS. For the convenience of users the names of its classes are close to traditional terminology and serve as one of the ways to address the system. At the same time based on the same data a comprehensive set of features was proposed for characterizing surface syntactic, semantic, pragmatic and discursive features of IS (see [6]) in order to permit in future to study their correlation with intonation parameters (focal accents, tone configurations, durations, tempo, phonations, prosodic timbres, pausation). At present the DB contains 264 files of dialogical IS utterances characterized for their prosodic, lexical, syntactic, semantic, pragmatic and discursive features. 2. Prosodic description of an IS in the DB Each file of the DB has undergone intonation analyses and as a result received a prosodic description in terms of prosodic markers earlier introduced by S.V.Kodzasov for describing intonation (see [7] and [8]). The distinctive feature of Kodzasov’s system is the consistent differentiation of accents and integral prosodies. More traditional description of Russian intonation are given in terms of a restricted number of standard intonation patterns that do not permit to reflect the multi-dimensional character of utterance intonation that in its own turn reflects various parameters of its meaning: illocutionary, modal, evidential etc. 2.1. Notational conventions for prosodic description In case a prosodic phrase (syntagm) has a salient marked integral prosodic feature the former is enclosed in square brackets. The closing brackets of such phrases are labeled with capital letters in order of their linear position in IS: […]A […]B. In case some integral prosodic feature has a scope over the boundaries of some smaller prosodic phrases this scope is marked by additional square brackets and has its own label: [[…]A […]B]C. Vocalic center of the syllable which bears tonal accent is marked by a number that follows the corresponding vowel. The numbers are sequenced in linear order. Asteriscs mark focal (loudness) accents: * for normal and ** for contrastive focal accent. These accents are determined aurally because their reliable phonetic correlates remain yet unknown. Normally focal accent appears in combination with tonal one. But each of them can appear independently. The lengthening of the vowel is marked with ~, and its shortening is with “. If a vowel is followed with a digit, then these signs precede the digit. 2.2. Representation of integral prosodic features In the DB the following integral prosodic parameters of phrases are distinguished and marked: tone register: A= h (high) / sh (super-high) / l (low), sl (super-low) / inc (inclination), dec (declination) tempo: A = f (fast) / s (slow) loudness: A = ld (loud) / lw (low) phonation: A= asp (aspiration) / crk (creaky voice) / ln (lenis) / fr (fortis) peculiar reduction (in everyday communication with “the likes of us”): A=rd 2.3. Representation of tonal accents Tonal (pitch) accents in the DB are characterized by marking the values of the following parameters: direction: 1=/ (rising), \ (falling), /\ (rising-falling), /= (rising- flat) etc. with possible “compositions” of tonal accents, e.g. 1-2=/=\ register: \l (falling in low register), /l (rising in low register),\h (falling in high register), /h (rising in high register); neutral register is not marked value of tone interval: 1=// or 1=\\ (extended interval of about 1.0 octave), 1=/” or 1=\” (small interval of about 3-4 semi-tones); neutral values (about 6-7 semi-tones) are not marked localization on the accentuated syllable: 1= \C (movement basically on the consonant), 1= \V (movement basically on the vowel), /C\V (bidirectional movement on the consonant and the vowel), 1=\NL (movement is not localized and is performed on a string of syllables); neutral localization of tone (movement within the non-final part of a syllable) is not marked co-phonation: tone movement may end with glottal stop (1=/q or 1=\q); have a creaky voice on the accentuated vowel(1=/crk or 1=\crk); end with aspiration or have it on an accentuated vowel (1=/asp or 1=\asp) 2.4. An example of prosodic indexation An example Each file in DB contains a dialogical turn that has at least one instantiation of an IS and has an audition button, that enables one to hear the recording. Access window for prosodic information has the following zones: 1) File address with 3 positions: q (question), number, m (musc.)/f (fem) 2) orthographic form of the IS; 3) prosodically indexed orthographic form of the IS; 4) deciphering of prosodic indexes for accents 5) deciphering of prosodic indexes for blocks 6) comments Thus, we have the following prosodic representation for one of the turns in our DB: 1) q22m 2) Вы сами придумали? Или это литературный сюжет? Вы *са1ми это придумали? || [Или это *литерату2рный сюжет?]A 4) 1 = //; 2 = \\ 5) A = inc The other 4 Access windows give detailed surface-syntactic, semantic, pragmatic and discursive information on each IS in a dialogical turn (see [6]). 3) 3. Conclusions Thanks to its multi-dimensional character DB “Intonation of Russian Dialog” may become a useful instrument for further study of intonation in its correlation with different parameters of utterance syntax and semantics. Indexation of DB items in terms of values of these parameters also provides the possibility to study their compatibility within IS and to choose their most frequent combinations as well as to determine intonation patterns that express semantic clusters. 4. Acknowledgements We would like to thank the RGNF foundation for their help in the form of the grant. 5. References [1] Pierrehumbert, J. B. The Phonology and Phonetics of English Intonation. MIT PhD Thesis. 1980-1988. [2] Beckman, M. E. Stress and Non-Stress Accent (Netherlands Phonetic Archives No. 7). Foris. (Second printing, 1992, by Walter de Gruyter.) [3] Selkirk, E. “Sentence Prosody: Intonation, Stress, and Phrasing.'' In John A. Goldsmith, ed., The Handbook of Phonological Theory, 1995, 550-569. [4] Русская грамматика. Т.2 . М., 1980 [5] Рестан, П. Синтаксис вопросительного предложения. Осло: Universitetsforlaget, 1968. [6] Kobozeva, I. M. An essay in characterizing lexicalsyntactic, semantic and pragmatic propertiea of interrogative dialogical turns in terms of features. Computational Linguistics and Intellectual Technologies. International Conference Dialogue’2005 Proceedings (Zvenigorod. June 17 2005) [7] Кодзасов, С. В. Комбинаторная модель фразовой просодии // Просодический строй русской речи. М., 1996. [8] Кодзасов, С. В. Уровни, единицы и процессы в интонации // Вопросы фонетики. Вып. 3. М., 1999.